Advanced heat transfer fluids for direct molten salt line-focusing CSP plants

Abstract

Concentrating solar power coupled to thermal energy storage (TES) is a vastly growing industrial process allowing for the generation of dispatchable and green electricity. This paper focuses on direct molten salt line-focusing technology using linear Fresnel and parabolic trough collector systems. Direct molten salt technology utilizes molten salt as heat transfer fluid in solar field and TES medium. Nitrate salts can be applied since they cover a wide temperature range. As storage medium Solar Salt, a binary ${\mathrm{NaNO}}_3-{\mathrm{KNO}}_3$ (60–40 wt%) mixture, is most commonly used but variations of this system have promising thermal properties in terms of a lower melting temperature to minimize the risk of undesired salt freezing events. These modified salts are typically ternary, ternary reciprocal or higher order systems formed by adding additional cations, anions or both. In this study five molten salt systems Solar Salt, HitecXL (CaKNa//NO3), LiNaK-Nitrate, Hitec (NaK//NO23) and CaLiNaK//NO23 are both investigated and critically reviewed. Their thermo-physical properties including phase diagrams, composition, melting ranges, melting temperature, minimum operation temperature, thermal stability, maximum operation temperature, density, heat capacity, thermal conductivity, viscosity and handling are evaluated and the most recommended values are discussed and highlighted. This review contributes to a better understanding of how the listed properties can be determined in terms of measurement conditions and provides temperature dependent data useful for future simulations of direct molten salt LF CSP plants.

Introduction

Concentrating Solar Power (CSP) plants convert sunlight into thermal energy by increasing the solar radiation flux density to achieve high temperatures and high energy conversion efficiencies. CSP plants have been and still are one of the most rapidly growing commercial technologies. The installation of power plants in the mid 80’s and 90’s in California with a total of 354 MW_el has been one of the key milestones. [1] At the time of writing this article, the electric capacity of CSP plants has been increasing to more than 5GW_el, worldwide and is expected to further increase by installations planned especially in the middle and far east.

Compared to other power plants the number of individual components is high, making CSP a complicated but versatile process. [2] The production of thermal energy makes it unique compared to the other renewable energy technologies and offers “intrinsic compatibility with thermal hybridization and thermal storage” as described at length elsewhere. [2] Thermal energy is generated by a solar collector system (e.g. a heliostat field) that reflects direct normal irradiation towards a thermal receiver system, where optical energy is converted to thermal energy. This energy can be used to drive a superheat steam cycle or directly be stored in a thermal energy storage unit. The different types of CSP configurations known to date, available types of thermal energy storage (TES) as well as the materials used during heat conversion and storage will be introduced hereafter. The main focus of this work is to review available and novel heat transfer fluids based on molten nitrate salts, their thermo-physical properties and finally recommend a list of values to be used for future reference. Readers interested in a deeper insight into optical principles, CSP applications and their value are referred to other publications, e.g. the overview of Gauché et al. [2] published recently.

A number of concentrator configurations, as shown in Fig. 1 is known to-date in CSP applications and two groups are differentiated: Line-focusing (LF) and point-focusing (PF) systems. Installed power plants of the first group are dominated by parabolic trough systems, which exhibit higher collector efficiencies than Fresnel systems, which in turn are expected to reduce capital costs. [3], [4], [5], [6] The parabolic trough concept uses parabolic mirrors to focus sunlight onto an evacuated tube receiver, with mirror and tube following the daily movement of the sun (see Fig. 2). Linear Fresnel systems use individual plane mirrors to focus sunlight on a (downward-facing) tube receiver.

Both systems have in common that the heat transfer fluid is pumped through a relatively large solar field before entering the steam generator or the storage unit. Currently, the 280 MW_el-Solana-, 250 MW_el-Mojave Solar- and the 250 MW_el-Genesis Solar plant are the largest operational parabolic trough plants world-wide (see e.g. the comprehensive review by Weinstein et al. [7]). A 360 MW_el trough-power plant is under construction in Chile with an estimated start-up by the end of 2017. Typical operating temperatures of LF plants are between 290–390 °C with peak and annual average conversion efficiencies of 14–20% and 13–15%, respectively [8].

To date, no large-scale commercial power plant utilizing direct molten salts storage is available to our best knowledge, but is under investigation in the CSP market (see also ref. [6], [9] and literature cited therein). In particular, the technology was thoroughly investigated by Sandia National Laboratories [10] and ENEA [11] in pilot scale plants using sodium and potassium nitrates (Solar Salt).

Central receiver systems are typically based on two-axial mirror (heliostats) systems that allow for high concentration ratios and higher temperatures in the receiver (solar tower or dish systems) compared to LF systems. Point focusing systems such as solar tower systems using molten salt as heat transfer fluid have first been demonstrated in the THEMIS project (France) in the 80’s and the Solar Two project (USA) in the 90’s (Fig. 2) as well as more recently on a relevant industrial scale such as the 110 MW_el-Crescent Dunes- or the 377 MW_el-Ivanpah power plants. In contrast to line-focusing systems high solar concentration ratios increase the maximum applicable operating temperature which benefits the power block, making this type of CSP-TES combination the most advanced at the moment. The high operation temperature between 290 and 565 °C result in higher peak (23–35%) and annual conversion efficiencies (14–18%) than LF plants [15]. This is especially related to the different power block fluid conditions and thus power cycle efficiencies (37.7% in LF vs. 41.6% in PF systems, see ref. [8] and literature cited therein).

The principle of concentrating and absorbing solar energy ultimately requires for a transportation of heat to a power block or heat exchanger. With the aim of achieving high operating temperatures, which are beneficial for the efficiency of the power block, demands for heat transfer fluids (HTFs) with wide operating-temperature ranges are created. At the same time, the type of solar concentrator used limits the choice of suitable HTFs. This issue becomes especially clear when looking at the existing technologies and HTFs used. Line-focusing CSP plants use thermal oils in a temperature range from below 0 °C up to 400 °C while power tower systems use molten nitrate salts operating between 290 °C and 565 °C. The reason for the different choice of HTFs is the design of the CSP unit itself. While in PF systems the HTF is pumped only from the tank into the close-by solar tower, in LF CSP systems it must be pumped through kilometer-long pipe networks of each solar collector assembly (e.g. $>50$km in SEGS I [16]). If the freezing point of the HTF is above ambient temperatures the risk of freezing events in the receiver tubes increases dramatically, especially whenever sunlight is absent. In PF CSP systems the number of absorber pipes is significantly lower and HTFs with high melting points can be utilized with a lower risk of failure and a reduced demand for preheating. Ultimately, these are the reasons why molten salts are not as easily applicable in LF systems while they are state-of-the-art in PF CSP plants. For LF CSP plants, trace heating or daily drainage of the receiver tubes is technically possible to avoid freezing events but not desired for LF systems with molten salt (e.g. cost of trace heating). Therefore significant effort is made to identify new molten nitrate salt mixtures with low melting points. Desired maximum temperatures for LF molten salt systems are close to the existing Solar Salt upper limit of 565 °C or lower. These optimum temperatures are defined by optical efficiencies of the solar field.

Thermal energy storage (TES) is a key element of CSP power plants since it is among few techniques offering dispatchable electricity in the MW-scale [17]. Moreover, no additional requirements are met for the existing electrical grid, making CSP-TES a perfectly integrated solution. General integration concepts have been proposed and are well summarized e.g. by Steinmann [18]: The storage unit is typically connected to the solar collector (solar tower, linear Fresnel, parabolic trough unit, etc.) [19] as well as to the thermal cycle (typically a steam turbine). Operation during sunlight allows for a direct operation of the thermal cycle, with surplus energy being stored in the TES unit. During discharging energy from the TES unit is used to heat the working fluid of the thermal cycle. Several advantages may arise from that:

• The capacity factor can be increased from 20 to 25% for CSP without TES, up to 60–85% for (PF) CSP coupled with a TES unit [20]

• Part load operation is reduced significantly

• Maximum revenue can be generated by shifting electricity-generation to “peak demand times”.

Developing a storage system requires for profound knowledge of the existing heat flows, between absorber and storage unit, as well as storage unit and thermal cycle during charging and discharging, respectively. A consideration of the different types of solar collectors, described before is essential herein.

Although different types of TES such as sensible heat in solids, latent heat and thermo-chemical storage exist [21], [22], [23], [24], sensible heat storage in liquids is most commonly applied in current CSP power plants. In CSP TES systems with thermal oil, molten salt and pressurized water were examined. Most importantly, this technology can be divided into direct or indirect storage technologies (see e.g. ref. [25] and literature cited therein) as shown in Fig. 3 both being typical options in point- and LF CSP plants, respectively.

Direct storage in mineral oil (e.g. Therminol VP-1 or Helisol® 5A) up to 300 °C has been applied e.g. in the SEGS-I power plant built in 1984 in California, using parabolic trough-shaped receivers [16]. It is possible to increase the temperature from 300 °C with mineral oil to 400 °C with synthetic oil. Costs of mineral oil could be acceptable whereas costs of synthetic oil are typically not acceptable for TES. Despite the advantage of having only a single circulating fluid which also serves as heat storage medium, drawbacks are the synthetic oil price, its thermal stability of up to about 400 °C and pressurized design above 300 °C as well as potential environmental hazards upon leakage and flammability and make this type of storage unattractive for nowadays applications.

To reduce the costs of the working fluids thermal oil is retained as HTF but replaced with molten salt as storage medium (typically Solar Salt - a 60–40 wt% ${\mathrm{NaNO}}_3-{\mathrm{KNO}}_3$ -mixture). This configuration is referred to as indirect storage configuration which requires for an additional heat exchanger between thermal oil (HTF) and molten salt (storage medium) as schematically shown in Fig. 3. This concept is still state-of-the-art and most commonly utilized in modern parabolic trough CSP plants. Yet, a complete replacement of thermal oil by Solar Salt is desirable for several reasons. The output temperature of the Solar Field can be increased to 450–500 °C which increases the efficiency of the power block (see e.g. González-Roubaud et al. [8] and literature cited therein). The successively higher ΔT increases the storage capacity (Q) per volume which is defined by

$$Q\left[\mathrm{kJ}\right]=m·c_p·\Delta T$$

where m in [kg] is the mass of the storage medium, c_p in [kJ/kg · K] is its heat capacity (typically constant over temperature) and ΔT is the applied temperature difference between the hot and cold tank. Accordingly, for a constant mass of storage material the storage capacity increases linearly with increasing ΔT; in other words for a constant storage capacity the required storage mass can be decreased. This allows for capital cost reduction of the thermal energy storage system. At the same time molten salts are significantly cheaper than synthetic thermal oils and a double investment of thermal oil as heat transfer fluid and molten salt as storage medium is avoided.

Overall, several authors showed that higher operation temperatures, larger temperature differences and replacement of oil of the direct molten salt system lead to reduced Levelized Costs of Electricity (LCOE) [2], [18], [26]. In addition molten salts have advantages in terms of handling (e.g. low environmental impact) compared to oil. Technically however, a complete replacement of thermal oil by molten nitrate salts resembles a significant challenge since molten salts exhibit a significantly higher melting point with the risk of salt freezing and/or filling/drainage of the solar field tubes compared to thermal oils (e.g. 240 °C for Solar Salt compared to −60 °C for Helisol® 5A [27].). At this point the operation principle of LF CSP plants is disadvantageous compared to PF ones and substantial effort is made in replacing thermal oils and eventually implementing molten salts as working fluids to increase the solar-to-electricity efficiency of the LF CSP plant. Additionally, different publications have proposed that investment costs can be reduced significantly when classical Solar Salt, a 60–40 wt% mixture of ${\mathrm{NaNO}}_3-{\mathrm{KNO}}_3,$ is replaced by molten salts with a lower melting point [24], [28], [29], [30], [31]. Kelly and Kearney [32] suggest that capacity specific costs for a direct molten salt storage system are around 50% lower compared to an indirect storage system due to reduced tank sizes and the absence of a costly heat exchanger, mainly.

Nitrate salts have been well known as heat transfer media in the processing industry [37], with the first established salt being Hitec HTS $({\mathrm{NaNO}}_2-{\mathrm{NaNO}}_3-{\mathrm{KNO}}_3,$ 40-7-53 wt%) developed by DuPont in 1938 and extensively characterized by Kirst et al. [35] in 1940. In the 1970’s and 1980’s the interest in molten salts for CSP applications grew dramatically. However, despite nitrate/nitrite salts being used in industrial processes to a significant extent, documentation of experiences has been scarce. Carling and Mar [37] were one of the first collecting data from production plants utilizing nitrates, nitrites or mixtures of both, as heat transfer fluids mainly. The widespread of applications ranges from organic syntheses, to sodium nitrate production, caustic concentrators and metal treatment. As a short summary, their discussion with the plant technicians demonstrated that no consensus existed on the treatment of the molten salts. For example, some plants used protective atmospheres for the storage of Hitec-salt, while others used a N₂/CO₂ gas mixture or atmospheric gas. Regeneration of molten salts was typically carried out by re-addition of the salt mixture or one of its components after defined time intervals. Continuous salt analysis was however carried out in all cases as a precaution, typically by measuring the melting point over time, or the salt composition and pH after dissolution in water to measure chemical changes in the salt. The experiences gained served as a basis for the planning of CSP plants (also described by Carling and Mar [37]) since molten salt treatment was considered inevitable.

Nowadays a binary mixture of ${\mathrm{NaNO}}_3-{\mathrm{KNO}}_3$ (60–40 wt%), known as Solar Salt, is the standard storage medium and serves as reference material in many publications regarding storage or heat transfer molten salts [38]. Its composition differs from the equimolar eutectic composition to reduce capital costs, since NaNO₃ is less costly than KNO₃ and NaNO₃ has a higher heat capacity compared to KNO₃. The effectively increased melting temperature, or more precisely liquidus temperature ($T_m=240$ °C), of Solar Salt compared to the eutectic mixture ($T_m=220$ °C) is considered tolerable.

The first large-scale power plants using indirect storage based on synthetic oil as HTF and Solar Salt as storage medium were the Andasol 1, 2 and 3 plants (all around 50 MW_el) which can be operated for more than 7 h by thermal energy provided by the indirect storage unit [39]. Operating temperatures of the 28500 tons tons of molten salt inventory (for each Andasol plant) are between 295 °C and 385 °C. At the time of writing this article, the largest parabolic trough power plant using indirect molten salt storage is the afore-mentioned Solana power plant in Arizona providing 6h of storage by 12 storage tanks comprising a total of 125 000 tons of molten salt.

Using direct molten salt storage and thereby increasing the maximum process temperature is one of the most promising options for CSP plants today. Increasing the hot tank temperature and therefore the specific storage capacity can significantly reduce tank volumes and the total required storage inventory. The Solar Two project has experimentally investigated the performance, stability and economics of a central receiver (35 MW_th) concept using molten salt as HTF and storage medium, connected to a 10 MW_el Rankine cycle. The integrated molten salt volume was 1400 tons (107 MWh, 3 h of autonomous operation) and thermal losses to the environment were in the range of 100 kW in the hot and 50 kW in the cold tank. Within the project salt chemistry changes were observed which however did not affect the performance of the system. Furthermore, no limitations deriving from corrosion phenomena could be observed after exposing construction materials to molten salt for up to 30000 h [13]. The first commercial power plant based on direct molten salt storage was the Gemasolar plant operational since 2011 generating 19.9 MW_el with a Solar Salt inventory of 7900 tons and a storage duration of 15 h. It is one of the first CSP plants that allows for a continuous 24-h operation.

Extensive effort has been put into the investigation of potential molten salt candidates for parabolic trough systems and most of them can be considered variations of Solar Salt [29], [40], [41], [42]. Those include the addition and/or exchange of cations and anions resulting in a vast number of potential salt systems and an even higher number of publications related to the identification of their thermo-physical properties. Unfortunately, there is no consensus on a significant number of these properties, making simulations of TES systems based on those parameters challenging and partly unreliable. The authors, which are a consortium of experts in the field of sensible heat storage, critically assessed and partially re-measured the most significant thermo-physical properties of a variety of molten salts intended for use as HTF and storage media in line-focusing direct molten salt storage plants.

This work summarizes molten salt properties relevant for direct molten salt technologies namely, liquidus temperatures, high temperature stability, density, specific heat capacity, viscosity and thermal conductivity. Finally, a list of recommendations will be given based on a critical literature survey and own measurements. The most relevant salt systems have been considered ranging from binary Solar Salt to ternary Ca,Na,K//NO3 (HitecXL) and Li,Na,K//NO3, to ternary reciprocal Na,K//NO23 (Hitec) and quinary reciprocal Ca,Li,Na,K//NO23 making this work a fundamental basis for modeling approaches not only, but particularly in the field of direct molten salt storage technologies.